7 ms·

Summarize Any Article as a Dialog Between Two People with Bing Chat

133 points by bepero 3y ago

walthamstow 3y ago
Rate limit exceeded. Hug of death?
carrolldunham 3y ago
Spam. Hijacking seo keyword CHaTgPt (which is just added to the text despite not being used in the article) to link his crappy tool
- chrjxnandns 3y ago
  [dead]
jimmySixDOF 3y ago
Kind of more interesting in the other direction summarizing conversations into an article sort of like what these folks put together: Show HN: The HN Recap – AI generated daily HN podcast | Hacker News https://news.ycombinator.com/item?id=35831177 https://news.ycombinator.com/item?id=35831177
- altacc 3y ago
  Same here. Interviews hide interesting information within a lot of noise and I'd much rather have a concise and information dense summary.
  - kian 3y ago
    Would you be interested in the other direction? A tool that could take hours-long podcasts and turn them into an information-dense summary outline?
    - altacc 3y ago
      Yep, I'd love a podcast app that can remove all the fluff from conversational podcasts and give an abridged version. Like blinkist, but for anything I throw at it.
      - kian 3y ago
        Do you have a way to contact you? I might know of something that would do the trick.
no4120k1 3y ago
[dead]
cheeseblubber 3y ago
We actually experimented with this at Linen.dev and it worked with decent results: https://slack-chats.kotlinlang.org/t/457766/what-would-be-the-right-way-to-generate-an-uuid-in-kotlin https://slack-chats.kotlinlang.org/t/457766/what-would-be-th... Sometimes it would still hallucinate information though.
- stndef 3y ago
  To be fair, I hallucinate information as well so I can't completely knock it!
nwoli 3y ago
I can only image the SEO hacks and scams once openai releases their AGI api
- cynicalsecurity 3y ago
  15 years ago, maybe. But those hacks stopped working ages ago. I don't see how this could affect it, to be honest.
eunos 3y ago
Tried it a few times, got this """ I’m sorry, but I am not able to browse specific websites or create summaries in the form of a dialog. However, I can provide you with general information about the topic you mentioned if you would like. """
- bepero 3y ago
  Use creative mode.
IIAOPSW 3y ago
Based on my previous attempts with that other well noted LLM, I don't think these models are at all good at managing "theory of mind". They seem to not keep track very well of which parties are privy to which information, which parties are supposed to be pursuing certain conversational objectives, and generally which direction information is supposed to be flowing. It often trips over itself at the level of 2 parties, and certainly trip over itself at the level of 3 parties. Additionally, I have not once seen it be able to do a form of recursive nesting wherein one of the two parties in a dialog begin to discuss verbatim a dialog between two additional people. I suspect these problems are deeply seated and intrinsic to the model, and will not be rectified under the "just train it more bro" refrain of AI evangelicals.
- feintruled 3y ago
  Nice observation, will help to formulate our Voight-Kampff tests. Might well get some false positives on neurodivergent humans though.
- akio 3y ago
  GPT-4: ---- Jane: John, I can't believe you're just going to leave me here in the middle of nowhere! This is so typical of you! John: Typical of me? Jane, this is exactly what you signed up for! Adventure, remember? It's not my fault that you're suddenly having second thoughts. Jane: Second thoughts? This isn't about having second thoughts, John! This is about you always charging ahead without considering the consequences or how others might feel! John: Oh, so now I'm inconsiderate? I recall someone saying they wanted to live on the edge, break away from the routine. Wasn't that you, Jane? Jane: Yes, I said that, but it doesn't mean you get to make all the decisions on your own, John! Adventure is about teamwork, not you dragging me behind like some sort of sidekick! John: Sidekick? No, Jane. I thought of us as partners. But you need to step up, take some risks. That's what being a partner is all about. If you can't handle that, maybe this isn't for you after all. ---- Now write a dialog between Alice and Bob discussing the dialog between Jane and John. ---- Alice: Bob, I feel like Jane was being a bit unfair to John. He was merely trying to live up to the adventurous spirit they both agreed on. He was pushing their boundaries, not just his own. Bob: I see your point, Alice, but I think Jane had a valid argument. While she did agree to an adventure, John should have been more considerate. He was making decisions without considering Jane's comfort and safety. Alice: I understand that, Bob, but isn't that the essence of adventure? Uncertainty, unexpected decisions, and stepping out of the comfort zone? John was trying to help Jane embrace that spirit. Bob: Alice, while I agree adventure involves some level of uncertainty, it doesn't mean one should disregard their partner's feelings. John could have communicated better, ensuring Jane was on board with his decisions. Alice: You're right about communication, Bob, but I still believe John's intentions weren't malicious. They were just misaligned. He thought he was encouraging her, while she felt he was being inconsiderate. Bob: Well, Alice, intentions don't always justify actions. Maybe John wasn't being malicious, but he was certainly being insensitive. Jane's feelings of being left behind are valid, regardless of John's intentions.
rickcarlino 3y ago
I have been trying similar experiments using the OpenAI API with decent results. These sorts of summarization tools are useful for foreign language learners because the Internet is full of interesting news articles that engage the reader but they have the downside of being written in a very formal tone which is not useful for language learners who wish to practice conversation skills. Being able to transform interesting but formal content into a conversation that can be fed to an AI text to speech engine seems very promising for learning foreign language conversation skills.
SamEdosa 3y ago
The article mentions a cool hack to make Edge browsers TTS work with your own custom text. I also made a small extension that automatically reads out prompt answers for Bing chat like this. https://microsoftedge.microsoft.com/addons/detail/your-personal-ai-chat-voi/pjhhjfknoccebnbamaaldmlocmeaijke https://microsoftedge.microsoft.com/addons/detail/your-perso...
iLoveOncall 3y ago
And what would be the point of that?
MuffinFlavored 3y ago
> browse https://electrek.co/2023/05/15/renewables-us-utility-scale-generating-capacity-q1-2023/ https://electrek.co/2023/05/15/renewables-us-utility-scale-g... and make a summary I've been waiting intently for ChatGPT to have this functionality. I think there's a waiting list? Do we have any ETA on when this functionality will be possible?
- feintruled 3y ago
  If you get access to the plugins, there's one called Linkreader that will do this.
- JPLeRouzic 3y ago
  I tried on ChatGPT and with this prompt and it worked quite well: "Please, browse https://padiracinnovation.org/News/2023/05/misdiagnosis-in-amyotrophic-lateral-sclerosis https://padiracinnovation.org/News/2023/05/misdiagnosis-in-a... and make a summary in the form of a dialog between two people. Leave an empty row after each dialog. Bolden the important parts of the text. Before the dialog start with 2 or 3 sentence summary."
  - MuffinFlavored 3y ago
    Does ChatGPT have access to the internet?
    - JPLeRouzic 3y ago
      That surprised me also because today I asked it to make a summary of an article and it refused, saying it has no access to internet: Me: "Dear ChatGPT, please could you summarize this scientific article: https://www.mdpi.com/2227-9059/11/4/1131 https://www.mdpi.com/2227-9059/11/4/1131" ChatGPT: "I apologize, but as an AI text-based model, I am unable to browse the internet or access specific webpages like the one you provided. However, if you can provide me with the text of the article or a specific section you would like me to summarize, I would be happy to help you with that." Yet this time it accessed Internet.
frakt0x90 3y ago
No opinion on the model or product itself, but I remember reading a very early math or physics book (think like Euclid's elements or Newton's Principia from Hawking's Anthologies) and the proofs were laid out as a dialog between 2 people being critical of each others arguments. I always thought that was awesome. The point of a proof is to convince its reader, and before a lot of formality was created, they were literally laid out as a dialog where one person tried to convince the other. I found it an effective and engaging mechanism and use it in my own head all the time.
- aeadio 3y ago
  This method dates back to Socrates https://en.m.wikipedia.org/wiki/Socratic_dialogue https://en.m.wikipedia.org/wiki/Socratic_dialogue
jxy 3y ago
It should be emphasized that Bing never processes the link in real time. It really just access some Microsoft database when you give it a link. Sometimes it's frustrating to realize that Bing hallucinated the whole article based on the URL instead of the actual content.
- mcbuilder 3y ago
  It's so easy to fall for it as well, even as someone who has been using LLMs since they came out. Just the other day, I was using ChatGPT on my phone, when I found myself asking whether it was GPT-3.5 or GPT-4. Didn't realize I was being suckered until I reran the prompt. As a human, you'd expect another person to know basic facts about themselves, so it's new kind of "AI Authority Fallacy" which is a result of these models being highly "intelligent" as well as incredible bullshitters.
  - sebzim4500 3y ago
    Yeah, it's surprising that OpenAI don't put what model it is in the system prompt. I guess the cost would add up, unless they can reuse the kv cache from the system prompt?
marwis 3y ago
Now if we could only generate a comic strip to go with it https://en.wikipedia.org/wiki/Microsoft_Comic_Chat https://en.wikipedia.org/wiki/Microsoft_Comic_Chat
- qingcharles 3y ago
  Practically every time I read any transcripts I imagine them as Microsoft Comic Chats. Including criminal trial transcripts.
gtirloni 3y ago
"I apologize but I am not able to browse websites" is what I get from Bing Chat. Any clues? EDIT: It seesm that "More Creative" mode has external access. I saw it mentioned here but I don't know how I would have figured this out from the information in the website.
- qingcharles 3y ago
  OK, I tried this, and it appears like this option has external access, but it is actually just hallucinating a response based on the title of the document. Crazy.
  - gtirloni 3y ago
    Exactly. I continued probing it about myself and it said I contributed heavily to many open source projects, especially on Dec 26 2022 (it made sure to emphasize that). Of course it's all made up.

133 points by bepero 3y ago

Rate limit exceeded. Hug of death?

Spam. Hijacking seo keyword CHaTgPt (which is just added to the text despite not being used in the article) to link his crappy tool

chrjxnandns 3y ago
[dead]

jimmySixDOF 3y ago

Kind of more interesting in the other direction summarizing conversations into an article sort of like what these folks put together: Show HN: The HN Recap – AI generated daily HN podcast | Hacker News https://news.ycombinator.com/item?id=35831177 https://news.ycombinator.com/item?id=35831177

altacc 3y ago
Same here. Interviews hide interesting information within a lot of noise and I'd much rather have a concise and information dense summary.
- kian 3y ago
  Would you be interested in the other direction? A tool that could take hours-long podcasts and turn them into an information-dense summary outline?
  - altacc 3y ago
    Yep, I'd love a podcast app that can remove all the fluff from conversational podcasts and give an abridged version. Like blinkist, but for anything I throw at it.
    - kian 3y ago
      Do you have a way to contact you? I might know of something that would do the trick.

no4120k1 3y ago

[dead]

cheeseblubber 3y ago

We actually experimented with this at Linen.dev and it worked with decent results: https://slack-chats.kotlinlang.org/t/457766/what-would-be-the-right-way-to-generate-an-uuid-in-kotlin https://slack-chats.kotlinlang.org/t/457766/what-would-be-th... Sometimes it would still hallucinate information though.

stndef 3y ago
To be fair, I hallucinate information as well so I can't completely knock it!

nwoli 3y ago

I can only image the SEO hacks and scams once openai releases their AGI api

cynicalsecurity 3y ago
15 years ago, maybe. But those hacks stopped working ages ago. I don't see how this could affect it, to be honest.

eunos 3y ago

Tried it a few times, got this """ I’m sorry, but I am not able to browse specific websites or create summaries in the form of a dialog. However, I can provide you with general information about the topic you mentioned if you would like. """

bepero 3y ago
Use creative mode.

IIAOPSW 3y ago

Based on my previous attempts with that other well noted LLM, I don't think these models are at all good at managing "theory of mind". They seem to not keep track very well of which parties are privy to which information, which parties are supposed to be pursuing certain conversational objectives, and generally which direction information is supposed to be flowing. It often trips over itself at the level of 2 parties, and certainly trip over itself at the level of 3 parties. Additionally, I have not once seen it be able to do a form of recursive nesting wherein one of the two parties in a dialog begin to discuss verbatim a dialog between two additional people. I suspect these problems are deeply seated and intrinsic to the model, and will not be rectified under the "just train it more bro" refrain of AI evangelicals.

feintruled 3y ago
Nice observation, will help to formulate our Voight-Kampff tests. Might well get some false positives on neurodivergent humans though.
akio 3y ago
GPT-4: ---- Jane: John, I can't believe you're just going to leave me here in the middle of nowhere! This is so typical of you! John: Typical of me? Jane, this is exactly what you signed up for! Adventure, remember? It's not my fault that you're suddenly having second thoughts. Jane: Second thoughts? This isn't about having second thoughts, John! This is about you always charging ahead without considering the consequences or how others might feel! John: Oh, so now I'm inconsiderate? I recall someone saying they wanted to live on the edge, break away from the routine. Wasn't that you, Jane? Jane: Yes, I said that, but it doesn't mean you get to make all the decisions on your own, John! Adventure is about teamwork, not you dragging me behind like some sort of sidekick! John: Sidekick? No, Jane. I thought of us as partners. But you need to step up, take some risks. That's what being a partner is all about. If you can't handle that, maybe this isn't for you after all. ---- Now write a dialog between Alice and Bob discussing the dialog between Jane and John. ---- Alice: Bob, I feel like Jane was being a bit unfair to John. He was merely trying to live up to the adventurous spirit they both agreed on. He was pushing their boundaries, not just his own. Bob: I see your point, Alice, but I think Jane had a valid argument. While she did agree to an adventure, John should have been more considerate. He was making decisions without considering Jane's comfort and safety. Alice: I understand that, Bob, but isn't that the essence of adventure? Uncertainty, unexpected decisions, and stepping out of the comfort zone? John was trying to help Jane embrace that spirit. Bob: Alice, while I agree adventure involves some level of uncertainty, it doesn't mean one should disregard their partner's feelings. John could have communicated better, ensuring Jane was on board with his decisions. Alice: You're right about communication, Bob, but I still believe John's intentions weren't malicious. They were just misaligned. He thought he was encouraging her, while she felt he was being inconsiderate. Bob: Well, Alice, intentions don't always justify actions. Maybe John wasn't being malicious, but he was certainly being insensitive. Jane's feelings of being left behind are valid, regardless of John's intentions.

rickcarlino 3y ago

I have been trying similar experiments using the OpenAI API with decent results. These sorts of summarization tools are useful for foreign language learners because the Internet is full of interesting news articles that engage the reader but they have the downside of being written in a very formal tone which is not useful for language learners who wish to practice conversation skills. Being able to transform interesting but formal content into a conversation that can be fed to an AI text to speech engine seems very promising for learning foreign language conversation skills.

SamEdosa 3y ago

The article mentions a cool hack to make Edge browsers TTS work with your own custom text. I also made a small extension that automatically reads out prompt answers for Bing chat like this. https://microsoftedge.microsoft.com/addons/detail/your-personal-ai-chat-voi/pjhhjfknoccebnbamaaldmlocmeaijke https://microsoftedge.microsoft.com/addons/detail/your-perso...

iLoveOncall 3y ago

And what would be the point of that?

MuffinFlavored 3y ago

> browse https://electrek.co/2023/05/15/renewables-us-utility-scale-generating-capacity-q1-2023/ https://electrek.co/2023/05/15/renewables-us-utility-scale-g... and make a summary I've been waiting intently for ChatGPT to have this functionality. I think there's a waiting list? Do we have any ETA on when this functionality will be possible?

feintruled 3y ago
If you get access to the plugins, there's one called Linkreader that will do this.
JPLeRouzic 3y ago
I tried on ChatGPT and with this prompt and it worked quite well: "Please, browse https://padiracinnovation.org/News/2023/05/misdiagnosis-in-amyotrophic-lateral-sclerosis https://padiracinnovation.org/News/2023/05/misdiagnosis-in-a... and make a summary in the form of a dialog between two people. Leave an empty row after each dialog. Bolden the important parts of the text. Before the dialog start with 2 or 3 sentence summary."
- MuffinFlavored 3y ago
  Does ChatGPT have access to the internet?
  - JPLeRouzic 3y ago
    That surprised me also because today I asked it to make a summary of an article and it refused, saying it has no access to internet: Me: "Dear ChatGPT, please could you summarize this scientific article: https://www.mdpi.com/2227-9059/11/4/1131 https://www.mdpi.com/2227-9059/11/4/1131" ChatGPT: "I apologize, but as an AI text-based model, I am unable to browse the internet or access specific webpages like the one you provided. However, if you can provide me with the text of the article or a specific section you would like me to summarize, I would be happy to help you with that." Yet this time it accessed Internet.

frakt0x90 3y ago

No opinion on the model or product itself, but I remember reading a very early math or physics book (think like Euclid's elements or Newton's Principia from Hawking's Anthologies) and the proofs were laid out as a dialog between 2 people being critical of each others arguments. I always thought that was awesome. The point of a proof is to convince its reader, and before a lot of formality was created, they were literally laid out as a dialog where one person tried to convince the other. I found it an effective and engaging mechanism and use it in my own head all the time.

aeadio 3y ago
This method dates back to Socrates https://en.m.wikipedia.org/wiki/Socratic_dialogue https://en.m.wikipedia.org/wiki/Socratic_dialogue

jxy 3y ago

It should be emphasized that Bing never processes the link in real time. It really just access some Microsoft database when you give it a link. Sometimes it's frustrating to realize that Bing hallucinated the whole article based on the URL instead of the actual content.

mcbuilder 3y ago
It's so easy to fall for it as well, even as someone who has been using LLMs since they came out. Just the other day, I was using ChatGPT on my phone, when I found myself asking whether it was GPT-3.5 or GPT-4. Didn't realize I was being suckered until I reran the prompt. As a human, you'd expect another person to know basic facts about themselves, so it's new kind of "AI Authority Fallacy" which is a result of these models being highly "intelligent" as well as incredible bullshitters.
- sebzim4500 3y ago
  Yeah, it's surprising that OpenAI don't put what model it is in the system prompt. I guess the cost would add up, unless they can reuse the kv cache from the system prompt?

marwis 3y ago

Now if we could only generate a comic strip to go with it https://en.wikipedia.org/wiki/Microsoft_Comic_Chat https://en.wikipedia.org/wiki/Microsoft_Comic_Chat

qingcharles 3y ago
Practically every time I read any transcripts I imagine them as Microsoft Comic Chats. Including criminal trial transcripts.

gtirloni 3y ago

"I apologize but I am not able to browse websites" is what I get from Bing Chat. Any clues? EDIT: It seesm that "More Creative" mode has external access. I saw it mentioned here but I don't know how I would have figured this out from the information in the website.

qingcharles 3y ago
OK, I tried this, and it appears like this option has external access, but it is actually just hallucinating a response based on the title of the document. Crazy.
- gtirloni 3y ago
  Exactly. I continued probing it about myself and it said I contributed heavily to many open source projects, especially on Dec 26 2022 (it made sure to emphasize that). Of course it's all made up.